Concerning trends and potential issues in osteosarcoma research publication

Osteosarcoma is a rare malignant bone tumor affecting approximately 5 per million persons per year in children and adolescents. Despite this rarity, there has been a substantial increase in the number of publications on osteosarcoma research, especially those involving cell-based experiments (Figure 1A). Given this discrepancy between the potential increase in knowledge about this malignancy and the impeded improvement in clinical outcomes,¹ we sought to understand the possible reasons for this difference and the trends in osteosarcoma research publications over the past decades.

We analyzed 416 basic research publications on human osteosarcoma (defined as papers with the term “osteosarcoma” in the title and the term “cell line” in the MeSH terms) published in 2020, around the year when the number of such publications peaked, to understand their content and identify potential issues. We found that publications from China (319 papers, 76.7%) far outnumbered those from other countries combined (97 papers, 23.3%), accounting for more than three-quarters of all publications this year (Figure 1B). As we went through these manuscripts, we noticed that a certain number of papers shared similar characteristics, including study design and figure layout and presentation. In addition to these subjective similarities, we also noticed more objective characteristics common to these papers. In particular, a significantly large proportion of the papers (41.1%) dealt with the potential functions of miRNAs, lncRNAs, or circRNAs in the pathology of osteosarcoma (Figure 1C). The overrepresentation of such papers (hereafter referred to as “miRNA/lncRNA” papers) is noteworthy because, although these molecules are known to play a critical role in some cancers, the evidence for the involvement of any particular miRNA or lncRNA in human osteosarcoma development is limited to date and is unlikely to represent the most critical avenues for current osteosarcoma research.²

Moreover, we found that these papers often used potentially problematic controls for the human osteosarcoma specimens and osteosarcoma cell lines. In particular, the use of “adjacent tissue” (also referred to as “normal tissue,” “para-tumor tissue,” “para-carcinoma tissue,” etc. in these papers) as a control for osteosarcoma specimens was found in 35.3% (147 out of 416) of the publications (Figure 1D). Because osteosarcoma patients typically undergo highly intensive preoperative chemotherapy, viable tumor specimens can only be obtained during biopsy. However, a biopsy is typically performed in a manner that minimizes contamination of surrounding tissues with tumor cells and pathologic fractures. Therefore, obtaining normal bone tissue at some distance from the lesion during biopsy is unfeasible and potentially unethical, as it can be detrimental to the patient. Bone tissue could be obtained during definitive surgery; however, gene expression patterns are likely to be significantly altered from normal conditions due to the effects of chemotherapy, immobilization, and inflammation. Additionally, the use of hFOB 1.19, a human fetal osteoblast cell line immortalized by transfection with a temperature-sensitive mutant of the SV40 large T antigen was found in 41.3% (172 of 416) of the publications (Figure 1D). Although mesenchymal stem cells or pre-osteoblasts are likely to be the origin of osteosarcoma, immortalized osteoblast cell lines, including hFOB 1.19, do not necessarily serve as a definitive normal counterpart for osteosarcoma cell lines.³ In addition, we found that 3.4% of the publications were retracted at the time of analysis.

The relatively high retraction rate of osteosarcoma “cell line” papers (retraction rate is less than 0.1% for biomedical journals in general⁴) and the concerning similarities found in the 2020 sample led us to investigate the characteristics of retracted osteosarcoma papers in the past. Retracted papers were identified using the query “(retraction OR retracted) osteosarcoma[Title].” This search yielded 404 results, and among them, we identified 229 individual papers after removing duplicates (mainly the retraction notices) and two case reports that happened to have the term “retracted” in their titles. We found that there was a sharp increase in the number of retractions after 2015, which followed the overall increase in the number of osteosarcoma papers with a lag of several years (Figure 2A). Among these publications, we were able to obtain 224 PDF files and review these publications. Of these 224 papers, 206 had at least one figure describing cell-based experiments, indicating that the majority of retracted papers was basic research papers.

While the specific reason for the retraction is not always explicitly stated on the journal website, we found that the concerning similarities noted in the 2020 sample were even more prevalent in the retracted papers. “miRNA/lncRNA” papers accounted for 67.0% (150 papers) of the retracted papers (Figure 2B), and the use of either or both “adjacent tissues” and/or the hFOB 1.19 cell line as a control accounted for 67.0% of all retracted papers (Figure 2C). Furthermore, similar to the 2020 sample, the use of these controls was significantly more common in “miRNA/lncRNA” papers (90.7%).

Taken together, these findings suggested that a significant proportion of the papers we reviewed had striking similarities in research topics, appearance, and even potential design flaws, and these characteristics were highly analogous to those found in retracted papers. This raises concerns about the validity of these publications and may suggest the possibility that at least some of them may have been produced from so-called paper mills using a template by someone unfamiliar with the intricacies of osteosarcoma research. A study by the Committee on Publication Ethics (https://publicationethics.org/) and the International Association of Scientific, Technical and Medical Publishers (https://www.stm-assoc.org/) suggests that potentially 2% of submitted papers may originate from paper mills,⁵ while another estimates a rate of 10.6 per 100,000 published articles.⁶ However, as our data suggest, these figures may underestimate the true prevalence of papers produced by paper mills.

Given the substantial increase in the number of publications on other cancer types, similar trends may exist in other areas of research. Furthermore, with the advent of generative artificial intelligence and its use, fraudulent papers will become, or may have already become, more sophisticated and much more difficult to detect. Therefore, not only publishers, but also researchers and reviewers, need to be aware of this potential problem and be more vigilant when evaluating manuscripts.

Keisuke Horiuchi: Conceptualization; data curation; formal analysis; investigation; methodology; writing – original draft; writing – review and editing. Kazuhiro Chiba: Supervision; writing – review and editing.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Approval of the research protocol by an Institutional Reviewer Board: N/A.

Informed Consent: N/A.

Registry and the Registration No. of the study/trial: N/A.

Animal Studies: N/A.